The amino acid sequence of a protein uniquely determines its tertiary structure. Deciphering this relationship, the protein folding problem has become increasingly important to molecular biologists. DNA sequencing has become routine, but structural experiments remain very difficult. Computational strategies are needed to help address this problem. This proposal describes a strategy to identify the location of alpha- helices and beta-strands throughout the sequence. A method for using off- lattice simulations of a polypeptide chain to identify secondary structure preferences in the ensemble average is proposed. Once secondary structure is located, computational methods exist for generating plausible tertiary structures. However, these combinatorial strategies give rise to a large number of alternative structures which are difficult to distinguish from the correct fold. Experimental and theoretical methods for clarifying the distinction between correctly folded structures and their misfolded counterparts will be considered. In a new direction, we propose to develop a multiple sequence analysis strategy to relate sequence and structure to function. In particular, we will focus on identifying the binding sites on the G-alpha family of GTPases for the relevant G-protein coupled receptors, G-beta-gamma and downstream effectors. We plan to continue to develop a genetic algorithm for the construction of polypeptide loops subject to a series of constraints. This method will be used to model the loop regions of G- protein coupled receptors involved in the interaction with peptide ligands and the hetero-trimeric G-protein complex. Finally, we propose to develop a new method to compare structures based on the area of the minimal "soap film" that could join them following the appropriate rotation and translation of one structure relative to another. This provides a natural way to circumvent the gap penalty problem that plagues current structure alignment algorithms.